Gresham College Lectures

What Is the Exposome and Why Does It Matter to Your Health?

November 08, 2022 Gresham College
Gresham College Lectures
What Is the Exposome and Why Does It Matter to Your Health?
Show Notes Transcript

Our health and susceptibility to disease are not wholly written in our genes. They are influenced throughout our lives by the environments in which we live, through our exposures to chemical agents, the infections we experience, to the psychosocial stresses of daily life. This appreciation of the role our environment plays in shaping our health and wellbeing is encompassed in the concept of the exposome, bringing together advanced statistical methods, exposure science and modern multi-omic techniques to better understand disease development and exacerbation. 


A lecture by Ian Mudway

The transcript and downloadable versions of the lecture are available from the Gresham College website: https://www.gresham.ac.uk/watch-now/exposome

Gresham College has offered free public lectures for over 400 years, thanks to the generosity of our supporters. There are currently over 2,500 lectures free to access. We believe that everyone should have the opportunity to learn from some of the greatest minds. To support Gresham's mission, please consider making a donation: https://gresham.ac.uk/support/

Website:  https://gresham.ac.uk
Twitter:  https://twitter.com/greshamcollege
Facebook: https://facebook.com/greshamcollege
Instagram: https://instagram.com/greshamcollege

- Well good evening everyone. Today, I am going to introduce a new ome, I'm going to introduce the exposome, but during that introduction, it reminded me of a talk, it must have been 15 years ago, when somebody said, we've always had an ome, and it's the most important one, it's econ-om-ics, and so we probably will need to focus on that a little bit as we go through this. But it's the exposome, and the science associated with it, expose-om-ics. But before I really delve into dictionary definitions, and there are, I'm afraid, I have to tell you, there are going to be jargon terms, and I'm going to try to explain them the best that I can as we go through this talk. I want to give you a sense that this topic has some gravitas, this is a topic about life and death, and the period in between. And I think that's important because it means that it's relevant to everybody, because I'm figuring everybody who's here, everybody online, let's give it, you've all been born, and let's face it, even with the wonders of modern medicine, at some point, each and every one of us are going to die. And that means that we really need to perhaps just pause a little bit and reflect on that period between those two extremes, it's the womb to the tomb concept. Specifically, this talk is going to be about how we age, how we develop disease, how we decline, and ultimately, yes, how we die, and how our environment, in its totality, across the entirety of our life course, contributes to those processes. Now, framed that way, this may sound as though it's going to be a pretty bleak 40 minutes, and so what I'm going to encourage you to do is to embrace the concept of memento mori, I want you to all embrace the concept and remember that, one day, we will die. And the reason I want you to reflect on that is it's only when you reflect on that that you can consider how we should live, and then, with a little bit, just a tiny bit of mental gymnastics, this doesn't need to be such a dismal, such a worrying presentation, because the exposome, and exposomics, becomes a science which isn't obsessed with identifying every individual chemical risk and hazard within our environment which can cause us harm it's actually more about identifying what is an optimal and equitable environment for the population to ensure that we live long and healthy lives, and that when we decline, instead of investing huge amounts of money in managing chronic diseases over decades, there's money available in our healthy age population to actually deal with decline and palliative care properly. This is about prevention of disease, that's what the science is really encouraging us to do. I can't duck dictionary definitions, but sometimes, they don't really help tremendously, and I'm going to spend some time on this one because this is a helpful definition of the exposome. It's taken from the 6th edition of "The Dictionary of Epidemiology," it's as good a source as any other,

but let's read it:

"A potential measure of the effects across the life course "of exposures and health," there you have it, the womb to tomb concept, "comprising the totality of exposures "that an individual is exposed to from conception to death, "including environmental agents," specific agents, such as chemicals, chemicals in the air, chemicals in the water, chemicals within our food, but also socioeconomic conditions that causes internalized psychological stress, aspects of our lifestyle, aspects of our diet, and then, on top of that, we have endogenous processes. Now look, there's a lot there, and underneath, we have this picture of the life course to illustrate that we're going to cover the whole caboodle, in a sense, but let's unpack this just a little bit. Life course exposures,

let's think about that:

every time we breathe, eat, drink, exercise, work, shower, sleep, we are exposed to cocktails and complexities of air pollutants in the air outside, synthetic chemicals in our air, pharmaceuticals that we take, UV, noise, the clouds of microorganisms that we navigate through as we move through our daily lives. And clearly, those exposures are going to be highly variable, not just across the period from the moment we're born until our late middle age, until we become elderly, to be quite frank, those exposures are going to vary between Monday, Tuesday, and Wednesday of a given week, and between the morning, the evening, the afternoon of a given day. So how on earth do we go about trying to capture the concept of the totality of exposure in the human population over that interval? And if you then just go, well that sounds like a big task so let's just park that over here for a second, then there's another issue, isn't there, 'cause the question is, we want to relate that to health, but what do we mean, and over what interval do we mean, what is going to be important? Are we really concerned about exposures in utero, or in those very early years of life, when organ systems are developing, and therefore individuals are vulnerable? Are we really more concerned about exposures that occur as we hit our late middle age, and our immune systems are slightly degraded, and we're not repairing damage in our body as well as we used to? Anybody over the age of 50 will know what that means. Or are we really talking about the sum of exposures over the entirety of a life course? And how do you measure that? And then what on earth do you relate it to? So far, I've talked about health, if you want to understand how our exposures cause disease, well it's a bit late if you take an exposure and you relate it to the presence of a disease, especially if that disease has killed you already. You kind of want to know how those exposures contribute towards that disease developing in the first place, and how that disease progresses, so in a sense, we want to move away from that terminal endpoint to understanding how the exposome, in its totality, plays a role in driving the evolution of chronic diseases. So that means we're looking for intermediate biomarkers on that path, because those biomarkers, potentially, will give us tools to survey the population to see how they're doing and to potentially intervene. So, an example, let's think of cardiovascular disease, let's park cardiovascular death, what we really want to know is, do our chemical exposures in our environment, are they linked to cholesterol in the blood, are they linked to arterial plaque formation, are they linked to high blood pressure, are they linked to inflammation, that's what we want to capture, and we want to understand that relationship on that pathway to disease. So that's quite a long intro, and I've just really been talking about life course exposures and health, and then we've got all this other stuff in orange here. What do we have? We have environmental agents, socioeconomic conditions, lifestyle, diet, endogenous factors. Okay, that's quite a lot to pile in to a consideration of health, and before we even do that, it should be apparent that lots of those things are going to overlap. I could stand up here and say, if I really wanted to study something which I know causes premature death and ill health, I could study poverty. If I focus on poverty, that's an amalgam of a whole combination of environmental stressors which cause harm to health, poor housing, pollution in the house, living in poor, polluted urban environments, poor diets, psychological stress from struggling to get by on a daily basis, all of those things compounding would contribute to those effects, so how do you unpick it, how do you unpick what's caused by poverty, and all of those factors, and the specific chemical agents within that scenario which would cause a consequence, a health effect? And then there's this, because you might conclude all we're really interested in doing is measuring chemicals in the water, chemicals in the air, and then relating that to health, but actually, all of those things have to be internalized, all of those things have to be metabolized in our body, and therefore our body changes the nature of those exposures and our body responds to those exposures. So we have an internal exposome which is influenced by our own metabolism, our own chemical transformation of those compounds, the interaction of those compounds with the bacteria in our intestines, with the bacteria in our lungs, all of which are going to change the nature of that exposure. I think I'm doing quite well here in piling on level of complexity on level of complexity, and I'm making these points now not to undermine the entire enterprise of understanding how the environment drives ill health in the population, but to give you a sense of the scale of the challenge, because I genuinely believe that understanding how our environment, and its interaction with our underlying genetic differences, interact to contribute to diseases, and significant global diseases, is one of the great challenges of 21st century science. It's a big one, a real challenge, a significant topic. So, the exposome, let's have some background. This is a faint and misty picture, cast your mind back, it's prior to August, 2005, and the exposome doesn't exist, it's not been mentioned, not in the way that I'm talking about at the present moment in time, not in this comprehensive, total life course concept of exposures to multiple things. It's not that we didn't appreciate that chemicals can cause harm, we have an occupational literature on metal fumes, on chemical vapors, on particles and fibers, and we know, and we've known for a very long time that those chemical exposures cause harm, are associated with a range of disease presentations. In fact, we can say that we've known that for quite a long time, the father of modern toxicology, Paracelsus, as he christened himself, wrote his first treatise on the diseases of miners, you can imagine the miner is technically exposed to a mining exposome, if you like, in 1533. So we know that, and let's face it, we knew from probably the 1950s that air pollution kills people, both in the short-term and the long-term, and in terms of cancer and chemically-induced carcinogenesis, we know the pathways, we know what makes those chemicals carcinogenic, we know the processes by which they cause DNA damage, we knew the processes by which they could promote tumor development, all of that was known, we had good biomarkers, so we're not starting from ground zero here, so the exposome's clearly something slightly different from all of that prior knowledge. But the reason that I have the exposome here, and I've knocked back that discussion of environmental factors, is because in the late '90s, into the 2000s, the environment wasn't where the action was, the action was in genomics, it was the golden age of the genomic revolution. And I think people genuinely believed that once we understood the variations in the human genome, we would have a key which would unlock our understanding of diseases. So if we cast our mind back to the Human Genome Project, it starts in America in around 1990, and by 2000, they have a preliminary draft, so we have our two chief architects here, we have Francis Collins and Craig Venter, representing the government and the independent business community's collaboration on this project, and they have a draft of the human genome. And by 2001, it's published for the first time, it appears in our most famous scientific journals, "Science" and "Nature," not complete, but complete enough, it takes 'til 2003 to get a much more complete version of the human genome, and then now we're in the post, potentially, genomic era, and the game changes slightly. We're less interested with the sequence of the human genome as an average, we're more interested in how it varies across individuals and across populations, we're become interested in subtle variations in the genetic code, SNPs. It doesn't stand for the Scottish National Party, it stands for single nucleotide polymorphisms, these variations in the gene which affect our genotype. And by identifying these, by sequencing large populations, we can begin to ask the question, how do genetic variations relate to disease traits, in genome-wide association studies. The next figure, and I know it's going to be very difficult, so I will describe it to you in great detail, is probably the figure which is most used in every talk which has ever been given on the exposome. And actually, what it shows is a series of traits. So here we have things like, at the bottom, we have eye color, hair curliness, but in here, also, we have diseases, stroke, cardiovascular disease, chronic obstructive pulmonary disease, we even have a whole range of cancers, lung cancer, colon cancer, bladder cancer, stomach cancer, for example. And what this axis is showing us here is the percentage of that trait which can be explained by the underlying genetic variation. And for some of them, it's a lot, eye color, definitely, type 1 diabetes, clearly, your genetics are largely driving those effects, but let's go up this list, if this is going from 0% to 100%, we get to about here, and suddenly, we're beginning to see that the genetics is explaining less than 50%, and we go up to here to cancers and we can see it's explaining maybe 5%, 10%, 2% of the variation. So the genetics isn't explaining all of this effect, we have this blue, amorphous area of, what is it, and what it is is the environment. This is an indication that it's not simply genes alone, it is the gene and the environment interaction. That's an interesting starting point, it's strange, in a sense, that this wasn't flagged up in a big way to begin with in a talk about exactly why that happens, but at this point in time, gene-environment interactions, it's not that people don't appreciate they're important, it's simply that, at this time, the G for Genomics is a capital G, and understanding environmental triggers is a very small e in terms of of the investment and the level of interest within the research community. A lot of these diseases are what we call non-communicable diseases, and they really, really matter. So let's begin. This is just a figure to give you a sense, it's taken from 2016. In 2016, non-communicable diseases were responsible for 41 million deaths globally. There are four major causes, the largest, of course, is cardiovascular disease, but here we can see cancer, respiratory diseases, when you have respiratory diseases here, it's normally talking about chronic obstructive pulmonary disease, bronchitis, emphysema type presentations, we have diabetes, and then we have other NCDs, which really is heavily dominated by dementia. These diseases are a huge global burden, and we know that most of that burden is caused by the environment. Now, it's not a completely blank slate, is it, because we could put our hands up and say, hang on, isn't this caused by cigarette smoking, isn't this caused by unhealthy diets and a lack of exercise, we could all come out with the lifestyle reasons for why we may see these increases, but it's not all of the effect, and that's kind of why we need to understand how the environment is actually contributing towards these effects, and there are other reasons, other reasons to consider this in great detail. Some of this is really self-evident,

but I'm going to say it anyway:

these diseases are diseases of age. What this graph shows is how cancers, cardiovascular disease, respiratory diseases all increase in the population as the population ages, and that is important because we have an aging population, not just in the United Kingdom and Europe, but globally, we have an aging population. As the population ages, therefore, we're going to see more of these diseases, not fewer of them. If we look at the global predictions for the population as we go from 1950, so these are historic, all the way to 2100, we can see not only the expansion of the population, but the fattening of this distribution, meaning we have more and more individuals of age. So if we have an epidemic, some people would describe it, of non-communicable disease, it's a problem which is going to get bigger over the next 50 years, it's not a problem which is going to get smaller, so we definitely need to understand this. And what does that mean? Well of course, it's fattening, we have many more elderly individuals, but actually, it means the population of vulnerable individuals is increasing across the board, children, I've included individuals of reproductive age, I think some people sometimes forget that people of reproductive age should probably be in a vulnerable group in terms of environmental exposures. So it's a big problem. But that's the global picture, how about in the United Kingdom? This feeds backs to the discussion I had about poverty, and how poverty captures a whole host of different environmental exposures, and what we have here are data from the ONS looking at life expectancy, and what they've done is they've taken the bottom 10%, the poorest 10% of the United Kingdom, and they've compared them against the wealthiest 10% of the population. And if you look at a male, on average, there's a 9.4 year difference in life expectancy. If you look at females, it's 7.7 years. That's the cost of not understanding the exposome, which is many things, but it's all captured here. If that figure is interesting, we can then begin to unpick a little bit, because within here we have another, I think even more important figure, because, actually, it's not the amount of life you have, it's the health you have when you're alive, and another way of looking at this is the years of healthy life, the years of life which are disability-free, and then the figures are, I think, jaw-dropping. For women, in the deprived areas versus the most affluent areas, the difference is 19.3 years, for men, 18.5, now, in the United Kingdom. I find that a remarkable statistic. And what's explaining the difference? Non-communicable disease risk. Again, the orange bars are the most deprived 10%, the teal colored bars are the most affluent 10%, and what we're looking at is age-standardized mortality rate per 100,000 individuals for cancer, circulatory disease, cardiovascular disease, dementia, in its broadest sense, and respiratory disease. And look at respiratory disease, it's over twice, it's a doubling of that effect. I picked those years deliberately so we weren't considering the additional impact of COVID, so this is the effect beyond that. Hopefully, I've made a case for why we should study NCDs, and why we should consider them in the United Kingdom, and why understanding what causes them is important. Now I'm going to throw in a few other things to demonstrate that the nature of our exposures have changed more in the last 150 years than we could ever have imagined. There have been significant changes globally, more people have moved from rural communities into cities. In 2007, for the first time in human history, more people lived in cities than in the countryside globally, and by 2050, it's estimated that 2/3 of the global population, 7 billion individuals, will live in cities, and that transition from a rural way of life to a city, with most of this occurring in low and middle income countries, is going to change completely the nature of the chemical exposures you're going to experience. And I've illustrated that here, this is a map simply where the height of these tall sort-of skyscrapers represent population density, and you can see the mega-cities, particularly the emerging mega-cities in Africa and the mega-cities across Southeast Asia and Central Asia, just how these populations are shifting. So that's a big general change in exposome, country to city. And there's another thing which has changed, and that's the nature of our chemical environment. Now, I pulled these figures out this week, so these figures are really very, very current, and I'm always surprised when I review them. We had an industrial revolution, we also had a chemical revolution, and since the 1800s, we have effectively produced over 200 million new chemical entities, over a relatively small period of human history. This isn't something we can evolve with, this is an explosion of new chemistry. If we simply focus, okay, on the chemicals which are licensed for use today, there are, and I can precisely say this, 359,206 as of today, at this recording period, fine. We have an organization in the United Kingdom, it's known as REACH. REACH stands, and I'm going to have look at this, I always get REACH mixed up when I'm going through this, so let me just have a quick look, REACH is our Registration, Evaluation, Authorization and Restrictions of Chemicals body. It checks that chemicals are safe. It's tested 7% of those chemicals which are available and in widespread use. So the chemosphere is there, we're surrounded by it, most of it, we are sitting with our fingers crossed, hoping it's not harmful, being optimistic, because, quite frankly, the people who study chemical toxicity cannot keep pace with the new chemical entities which are being produced and entered into our environment. And now I'm going to just make it a bit worse by saying that's the tip of the iceberg, 'cause those are just the actual chemicals when they're made, but if those chemicals go through us, or through livestock and animals, our body will change them through our xenobiotic metabolism, there'll be new chemicals which will emerge, and they will enter our environment, into our soil, into our rivers, where they will be modified by abiotic processes, processes not to do with life, sunlight, oxidation, but they will also be processed through the microbiota, the bacteria and the fungi within our environment, to create a multitude of new compounds unknown to man. So a lot has changed, we are surrounded by an entirely new chemical environment. And when I talk about this, we're talking about the usual suspects, aren't we, pesticides, plasticizers, phthalates, heavy metals, pick your favorites, I'll say micro-plastics just to keep some people happy, some nano-particles, we could do that as well, pharmaceuticals, this is our chemical reality. All of this is background, and it's background to this. Hopefully, I've made a case that we should probably spend some time understanding how the environment impacts on our health, and as I said, there wasn't the exposome prior to August, 2005, and then there was the exposome, in this editorial, published in "Cancer Epidemiology Biomarkers and Prevention" in 2005, by Christopher Wild. And it's an editorial, and it's really incredibly important editorial, because all he simply does is make the case for an equal investment in understanding how the environment contributes to chronic disease relative to the investment in the underlying genetic causes of disease. And he's motivated to do this because there are lots of things going on, which means that, potentially for the first time, we could do this in a substantial way. The first of this is to do with genomics. Large numbers of individual genomes are being sequenced, so we know more about genetic variation than we've ever known before. We also have very, very large new cohorts of human subjects being generated, so these are research studies which recruit large numbers of individuals into them, and those individuals are being genotyped, there's linkage to their health record, and there are the provision, I mean we call them bio-samples, but actually, usually, we mean blood and urine samples, taken from those individuals, stored, so people can do chemical analysis. We also have new technologies, we talked about it when I was introduced, we can now, at this point, look at the expression of every one of the genes simultaneously within the human genome using the technology of transcriptomics, we can look at the expression of all the proteins in a cell in the blood, proteomics, we can look at the metabolites in our blood and in our urine, reflecting the total biochemistry of the whole organism by a process called metabolomics, which allows us, now, to be able to look at individuals, look at what they're exposed to, and look at the underlying chemistry which is occurring in their body. But there's another thing which is happening which is incredibly important, and that is, none of this is easy. If you want to do a gene-environment interaction study, it's very difficult to do it unless you have tens or hundreds of thousands of individuals. If you're then looking at genetic variation, if you're then looking at every transcript in the genome, every protein, every metabolite, well, do you know what? This is the definition of big data, you need brand new computing resources, you need new statistical tools, you need bioinformatics, or you ain't going anywhere, you're just collecting a mass of data you'll never be able to analyze, And it's at this point that it all coalesces. This paper, and I reread it in great detail in preparing for this lecture, I'd have to say, you read it, it's still current, it's still prescient, and so what's absolutely remarkable is that this went without a trace. There was no fanfare, for five years, it barely registered a dot, it really didn't change anything very much, until 2010. And this the point in which it begins to accelerate away, and as always, it accelerates because it gets talked about in a more prestigious journal, more people see it, more people respond to it. And this is another editorial, so this starts with people just pontificating, but this is published in 2010, the two lead authors play an incredible role in kicking this field forward, Steve Rappaport and Martyn Smith, both at the University of California in Berkeley. And they, in this editorial, didn't say we need to do this, what they did is they said this is how it should be done. And they proposed two strategies, one which is defined as bottom-up, and the other one, which is called top-down, and they're really quite simple. One of them are measurements seeking a disease, and one of them is diseases, seeking a cause. So in the bottom-up approach, you attempt to measure all of the chemicals within the environment, within the air, within the water, and then you try to relate it to disease presentations to see if there are associations. In the top-down approach, you usually start with a disease group, and you get a blood sample, and you get a urine sample, and then you analyze it for practically everything you can see within that sample. You can be a bit more focused, so if we have a range of exposures here, we've got radiation, stress, lifestyle, infections, and so forth, you can take the exposome, all of that stuff in environment, but you can actually focus down your considerations in that biological sample on things which you know are potentially important, xenobiotics, heavy metals, for example. And you can also, at the same time that you're measuring chemicals in the blood, by definition, what falls out of those analyses are biomarkers of how the individual is responding: is there damage, is there organ damage, is there inflammation. So you automatically have a disease, measurements of the diversity of chemistry that you can see within the blood or the urine, and you have a biological response signal within that. Now, that sounds like it's crazy, can you imagine how difficult that would be? But I'm going to tell you now that the National Phenome Centre within the United Kingdom, the US National Environmental Health Sciences, have methodologies now where you can take blood, and you can measure tens of thousands of individual chemicals in them simultaneously, well validated, well purified, by proton NMR to look at structures, or by techniques such a mass spectroscopy. The technology is there to do this, this is live, for the first time, it can be done. Here, I just really wanted to illustrate it a bit more here and just give you a bit of a flavor. This is the concept, here we have our life course, from tomb to not quite womb, but assuredly heading in the general direction, head down, certainly, and here we have our bottom-up approach, and here we have measurements of stress, diet, pollution, chemicals, a whole host of things we can look at here and relate to the health of that general population. And over here, we have a disease and a biological sample, and in that biological sample, we can measure transcriptomics, we can look at proteomics, we can look at a whole host of things, including biomarkers which we know may be important in the disease. Now, when all this started, people would look at that blood sample and they would potentially do transcriptomes, proteomes, metabolome, all of that variation, even something known as the epigenome, which is essentially modifications to your DNA which affect the way in which it's expressed in the body, and they would do these one at a time because doing them one at a time is one hell of an undertaking statistically, bioinformatically, but now, increasingly, they're doing them all simultaneously. And what I have down here looks like a piece of modern art in a sense, and many of these papers do look like modern art, but they take all of the data, and then they see how it clusters together. And what you're looking for is coherence between the transcriptome, and the proteome, and also seeing how your disease, and the chemical exposures map onto those clusters to try to learn something about the causation of that disease. Those methods are often described as being agnostic, a strange word, it means that you do the study and you're blind, it's observational, you see what you see. You can describe as being hypothesis-generating not hypothesis-led. But that doesn't actually need to be the case, and there are halfway houses which are very useful because they allow you to get over some of the statistical problems with handling such difficult and large datasets. Science works by probability, we often say, are we certain we would see this 95% of the time versus 5% of the time? Well, if you look at 100 things, you see five things by chance. If you look at 100,000 things, if you look at 200,000 things, how many random observations do you get? So you really have to try to be parsimonious with how you look at this data. And there is maybe a framework by which we can actually use prior knowledge, and I'm a great advocate of this. I don't think we should have a new science and forget that there are 200 years of toxicology before it, or that we actually understand some of the fundamental mechanisms by which diseases are presented and they evolve. And we can learn from this kind of pathway, and it's known as an adverse outcome pathway, and the prototypical one here is to do with cancer. We know what the key characteristics of carcinogens are, we know they're typically electrophiles, or they are metabolized to electrophiles, which will bind to DNA, and it will cause damage, and it will impair DNA repair, we know lots about them. We know the hallmarks of cancer which contribute in the disease pathway to cancer evolution, they are things like evading death, being super proliferative, stimulating angiogenesis, we know that these are processes which have to occur, and we know that the end product of those things is the presentation of an overt cancer. That trigger, the molecular initiating event, the key events between that and disease, with these relationships here, can be mapped to other diseases to give a structure in how we interpret these very large exposomic studies. Fortunately, just over a year ago, this paper was published, and it's incredible timely, because it allow us to fill in those adverse outcome pathways for environmental agents. It was published by three superstars of the field, Annette Peters, Tim Nawrot, and Andrea Baccarelli, and essentially, what it does is it talks about what the hallmarks, what are the mechanisms by which environmental factors are likely to perturb a system which is likely to bring about a health endpoint. And some of them, if you're an old scientist like me, I would have said, yeah, knew that. Do they cause oxidative stress? I think people know what oxidative stress is, too many dangerous oxidations in the body overwhelming our anti-ox defenses, and associated with that, do you get inflammation? Can you show mutations, is it genotoxic? Then we have epigenetic alterations, does it change the expression of our DNA, mitochondrial dysfunction, endocrine disruption, altered cell communication, altered microbial communities, impaired nervous system. These were their hallmarks. Now, when I read them, one thing immediately jumped out at me, in that many of these things are associated with aging. So we can go through here and we can pick many aging-related things, including oxidative stress, and of course we know that NCDs are associated with aging itself, and it immediately suggests, therefore, that accelerating aging itself might be a biological pathway by which our environment causes us to develop NCDs. So, trying to turn this into a good metaphor, imagine you're a 53 year-old, jaded biochemist, who's happened to have lived in Central London for 25 years, I may be, chronologically, a 53 year-old, I suspect my cardiovascular system is probably around 60 years old because of the nature of the urban environment and the exposures that I've been exposed to, plus the fact I don't get to do much exercise, and I probably don't have the best diet in the entire world, so there you go. Do the combination of all these exposures drive forward all of these effects? And once you accept that, you can begin to look in different organ systems, and begin to think about what your intermediate biomarkers on that causal pathway between exposure and disease might look like. In the heart, we may be looking for cell death, inflammation, again, mitochondrial dysfunction, in the lung, we're looking at inflammation, cell death, epigenetic changes, changes in immunomodulation of the airway, so we can being to actually flesh this out a little bit. Now, all of that sounds very persuasive, but there are complexities, and I thought I'd pick an example to try to unpick this, to try to show you how difficult it would be. And so I'm going to do an adverse outcome pathway, and I want to focus on NO2, and I picked it deliberately because you're all here, which means that, arriving here, you've already breathed a huge, unhealthy level of nitrogen dioxide just as you walked along Chancery Lane, that's just life. So here we have NO2, and let's assume, over here, we have an acute endpoint, and this is important because, so far, I've talked about NCDs, and that's long-term consequences of exposure, and actually, lots of toxicology really focuses on what happens in the short-term. Here we have NO2, it's a strong oxidant, it's a strong nitrating agent, so its molecular initiating event is oxidation, driving, if you like, inflammation. But we're all different, and so the way we absorb NO2, the way in which the products of NO2 oxidation in our body are distributed, the way they're metabolized and potentially eliminated are going to vary for all of us, that would be some underlying genetic variation. But ultimately, we get an acute endpoint, and in NO2, these are usually exacerbations of preexisting disease, such as an asthma exacerbation, for example. But we breathe this every single day, and so, actually, it's not a single day, it's a persistent irritation over a long period of time. And the molecular initiating event's not going to change, it's going to still be an oxidizing and nitrating agent, but that constant irritation is going to cause the airway to subtly adapt, we're going to have airway remodeling, and associated with that, we'll have a loss of immune function, we'll have changes in lung volume. But again, what we're looking at is persistent, so the actual pathway and the endpoints begin to change. But it's more complicated than that again, 'cause nobody just breathes NO2 as you're walking along Chancery Lane, you're breathing NO2 and the particles coming from the exhausts of the diesel buses and the diesel gases, and they go up and down together 'cause they've got a common source, so they're very difficult to disentangle from a health point of view, and that sits within a cloud of other components within the particles in the air which are formed by secondary processes in the atmosphere, secondary inorganic, secondary organic materials. And so, in reality, what we have over the long-term is kind of a storm, multiple agents, each with variable molecular initiating events, all handled differently by the tissue in the body in terms of their absorption, distribution, metabolism, and elimination, and all of them acting concertedly over numerous ways to bring about their health effects. So that might be quite daunting, I'm going to say, really, you could have lots of question marks about the combined mixture effect, but there are some commonalities which map back to those hallmarks of environmental disease: you always get oxidative stress, you always get inflammation, you always get tissue damage. And I can illustrate how this can be stitched together, using the exposome concept, with reference to one of the studies which was recently performed, which adopts a process called meet in the middle.

Meet in the middle simply works like this:

if I know that A causes B, if I know that an environmental exposure causes a disease, great. If there are samples from that population before those individuals got the disease, and I can relate the environmental exposure to changes in those samples before they have disease, and those changes I identify can then be mapped to the disease they subsequently presented, I can triangulate, and that gives me a causal chain. This is a study which was conducted on a small subset of individuals within a large cohort known as EPIC, in Italy, and in this study, they had already demonstrated, very strongly, that air pollution exposures, to NO2, PM2.5, and PM10, was associated with a risk of cardiovascular heart disease. But they had plasma samples stored in one of those bio-banks in a big cohort, from 17 years before the earliest presentation of cardiovascular disease, and so they were able to go back and analyze that plasma with proteomics to look at all the proteins, and isolate DNA and look at changes in the methylation status to see if any of the genes and pathways were perturbed. And when they did that, they found that air pollution was associated with changes in the expression of a cytokine which drives very aggressive inflammation known as interleukin 17, and that there are changes in pathways, which could be seen through methylation, to do with oxidative stress and antioxidant responses. And certainly, interleukin 17 and oxidative stress, if you looked at them in isolation in the individuals who had cardiovascular disease, were significantly associated, so you get a loop. Air pollution causes cardiovascular disease, the roots of cardiovascular disease in that group are caused by air pollution causing oxidative stress and inflammation over many years within that cohort, that's the way the logic goes. But that's still putting an exposure on those individuals, and those exposures are modeled, you say, how was he exposed to air pollution? He lives here, I have a map, we've estimated it. We could do better than that potentially, because we can give people devices which allow us to look at their personal external exposome. This is a study performed by my colleagues Ben Barratt and Jenny Quint, and this is an air pollution monitor which you can carry on your shoulder. And in this study, it was carried by individuals who had COPD for months, recording air pollution changes, temperature and noise changes day by day over an extended period of time. And because it's a cheaper monitor, you have to validate that it works and that it performs as well as the best, most expensive equipment, and that's what being shown there. This is representative data. In this study, noise, temperature, carbon monoxide, PM1, PM2.5, particulate matter, particles in the air, but also you have recorded symptoms in these individuals, peak flow measurements, whether they where wheezy, whether they were breathless, did they have cough, did they produce sputum, symptoms of COPD exacerbation. And when you look at this data, the actual measurement reflects the individual, it isn't just an estimate of where they lived, and let's face it, nobody spends 24 hours a day, 7 days a week sitting on their doorstep. Again what you see is, this is NO2, here, we're looking at exacerbation frequency, and exacerbations are particularly harsh for individuals with COPD, we can see the NO2 on the day was associated with an exacerbation, but actually, it persisted for about three days after the effect. So having that measurement, which is your own personal assessment of the NO2 you were breathing, which follows you around outdoors, indoors, in the train, going backwards, helps. And in fact, you can see something of symptoms here. So again, we're looking at symptoms, we're looking for effects which are above this central line to see if they're significant, and we can see there are associations with cough and some associations with sputum production, a bit borderline, but you're looking at hundreds of subjects here, not tens of thousands. So personal monitoring is another way we can improve that understanding of the external exposome. But we do have a fundamental problem, which is, that doesn't map to thousands of individuals, and our bio-banks have thousands of individuals, and in my lifetime, I don't think we're going to be in a situation where everybody has a small device which measures the air pollution that we breathe, so modeling does matter. And there are different types of models you can use, this is a model which uses measurements and measurement campaigns, and then relates those measurements of pollution to geographical features, land use within cities, creates a regression model, produces a map. And over here, this is a different type of map which looks at the pollution from all sources in the air and then deals with dispersion, deals with meteorology, deals with chemistry. And this model, which was produced by a member of our team, Sean Beevers, actually gives you 20 meters by 20 meters resolution, minute by minute, so you can begin to actually do really good exposures at residential address, at post code, but it's not as good as personal monitoring, because it's not capturing the totality of exposure. But perhaps you can do something about that: this is the flux, there's a little clock here, starting in the morning, going through to the evening, this is the flux of people coming into London in the morning, milling around in London during the day, and going home. You'll see it pulse in the morning and then pulse at about five, six o'clock, seven o'clock as everybody leaves. It's derived from Transport for London's Travel Demand Survey, and it represents a snapshot of how people move throughout the city. So it's not everybody, but if you have a questionnaire, and you can see that some people behave in a particular way, you can begin to use this information to refine those exposure estimates. You can say, people spend, on average, eight hours outside in London, they spend, on average, 50 minutes on the Tube, maybe they take a car, maybe they take a taxi, and you can maybe do measurements on the Tube to find out how much pollution is in the Tube, and you can do measurements as people cycle, or measurements in the vehicles as people drive to work, and you can even do measurements in the home, where the exposures are going to be very different from outside, to get a totality of exposure. And then you don't treat an individual as an individual with a monitor, you treat human beings like ants, who you can cluster together into groups because, generally speaking, they do the same thing every day when they're at certain stages of their life, and you can begin to refine the models, and that's another way of trying to improve this estimate of the external exposome. I'm using air pollution, I'm not apologizing, I'm an air pollution person, but these approaches can be rolled out for other chemicals in the environment. And again, we get back to this, the exposome can help us here as well. I told you it's very complicated and we can't unpick things, so maybe, let's think about that. We can improve our exposure estimates, but, do you know, the most, I want to say, the best monitor we have of air pollution, or any chemical challenge, is the human, the human is the best sensor. If we can capture measurements within that individual which tell us something about their exposure. And I'll give you an illustration of how this can help unpick that difficulty when things go up and down together in complexity. This is a study which was performed in which individuals with ischemic heart disease, COPD, healthy individuals, were exposed in London, twice. Once, they walked up and down Oxford Street, because it's a ginormous outdoor diesel chamber, effectively, and then, once, they walked around the Round Pond in Hyde Park, lower exposure, further away from the traffic. And they took blood samples at various points in that study, and in that blood, they sequenced the blood to look at circulating RNAs, these are species which are in your blood which are potentially useful as biomarkers, they tell us something about whether organ systems are under stress. And what they found was some of them went up, some of them went down, go figure, but some of them were associated with adverse effects in particular tissues. So they saw changes in certain small circulating RNAs which are related to adverse effects in the heart, the brain, the pancreas, so you can see that the air pollution was having an effect beyond the lung in these individuals, very subtly. But that's not the most important bit, the next picture's a Venn diagram, and you'll remember them at school, they were very simple, this just looks prettier. And what you have here are different pollutants which were measured on those individuals by personal monitoring as they did the walks. PM2.5 particles, less than 2.5 microns, black carbon, the small elemental carbon particles you get from exhausts, PM10, PM2.5 with other bits from road wear, tire wear, brake wear, NO2, we've talked about that a lot, emitted from diesel engines, and ultra-fine particles, particles less than 100 nanometers. If I was putting these together and saying, does that cause an ill effect on a population, it would show that they all did and it would be impossible for me to pull them apart, but what's important here is when you look at the association between these pollutants and the circulating small RNAs, they're completely different. And this is telling us that the internal exposome is discriminating between these components which are correlated in the air, and giving us information to begin to pull apart their relative contribution to health. And this matters, understanding which bits of air pollution are harmful is going to be a really important thing to do, 'cause, currently, we regulate air pollution in the air based on simply how much stuff is floating in the air, which is a bit strange, if you think about it, in this day and age. So I've got some challenges here, I'm going to go through these quickly, in terms of taking this forward. Do you know what? We really need to do more measurements of contamination, and we need to make sure that our measurements work and that we can validate them. There's a thing here called measurement error, measurement error is just a fancy way of saying that, actually, we're not very good at estimating individuals' exposure. We need to think about mixtures and how we unpick them, we need to understand dose-response relationships, not just whether something is bad, we need to understand the shape of that curve, so we work out if we reduce X by so much, does it produce a health benefit, that really is important. We need to do more of this meet in the middle type science to understand how environmental exposures cause disease, early, not once the disease is established, and we need better methods, better validation, and validation of what we find.

But there are some holy grails here:

we really need some longer term biomarkers of exposure, we really don't have them, to be quite honest, our biomarkers reflect maybe a day. We can measure adducts in blood, that's albumin with covalent modifications of chemicals attached to it, that maybe gives us an insight of about a month, we can look at epigenetics, different patterns in DNA from methylation, that might give us some help, we can look at biological aging, I think there's a lot to be done in that area, and we can look at accumulative tissues, like bone, through imaging, we can look in bone, and see, perhaps, metal accumulation, we could look in lymph nodes, which accumulate material in the body, but I would included here hair, one centimeter per month, the chronology of your chemical exposures locked in the cortex of your hair, teeth, there are archives of babies' milk teeth that can be used to look at chemical exposures in those children, and I'm going to say it, because they do exist, there are toenail clipping archives, where you can look at chemical accumulation in toenails. The other good thing is that we have bio-banks, and we have big bio-banks with resources which would allow us to do this, and as of this week, they are recruiting for this new cohort study, I'll have to read this correctly, Our Future Health. The UK Biobank was half a million, they want to recruit 5 million individuals, genotype information, linkage to health records, bio-samples collected through time to look at early biomarkers of disease, fantastic opportunity to really expand research in this area. And with that, I'm going to come to the end, but I just wanted to end by saying something, just trying to pull things together a little bit. I think the exposome has arrived. Some science has been done, I've scratched the surface, it's really beginning, it's beginning to take off, and it's going to be incredibly important in understanding serious disease risks in our population. There are challenges methodologically, it's challenging in terms of scale, we need long-term biomarkers, but I wanted to just leave you with one final message to illustrate why this matters. The United Kingdom is the fifth, depends on the day of the week, it's the sixth largest economy in the world, and there's almost 20 years of difference in healthy life between the poorest 10% and the most affluent 10%, and that reflects the totality of their lived experience, their lifestyles and their environmental exposures, and so the exposome really matters to them, it should matter to everybody who has any policy or responsibility for improving public health in this country, even if, until today, they never knew the word existed. Thank you. (audience applauds) - [Questioner] I'm curious as to how you manage the policy engagement, because you cut across, clearly, so many departments, how do you approach it, how does it work today, and how do you think it should work in future to be effective? - I can answer the question not with a complete exposome hat on, but I can talk about air pollution, and how you can do policy in air pollution, and I think it works two ways. The boring stuff is you publish papers, it gets peer reviewed, you sit on government committees, you give them information, and that is part of the job, but if you really want to shift policy, you have to educate the public. The public education is really important, because, to be quite frank, it's the public who motivate politicians to make change, the scientists just create the background, and so I think that's really where the critical issue is, I think people do need to know this. But, I mean, how many people in this room knew the 20 years of healthy life difference between the poorest and the wealthiest person in the United Kingdom? I could argue the best way to resolve that inequity is if we gave everything which belongs to the 10% of the wealthiest people to the poorest people, we would gain that almost 20 years of healthy life back, so we've got some big issues we have to deal with. - It's a very interesting question, because we have a lot of people with PPE degrees running our governments, and very few with science degrees running out governments, do you think there's an input error that we could correct by altering the type of people who govern us? - We had a lecture by Patrick Vallance in our center recently, and in it, he made the observation that when he took over his role as the chief scientific advisor, he did an audit of scientific advice throughout the departments of government, and unsurprisingly, he realized that there were whole departments which were a desert in terms of scientific advice because they're largely populated by humanity students or people with PPE, as you say, and so part of his remit, one of the things he was driving forward, was trying to improve, if you like, the scientific engagement with Parliament. And there are people doing it from the outside as well, there's an organization called Sense for Science, who have regular evidence weeks, they've had one up in Scotland, in Holyrood, they have them coming up in London, where they get scientists to actually go to the chambers and actually be there so that the politicians can go up to them and ask them questions about evidence, and that personal contact, I think, is critically important. But you're right, there is a deficit within government. And the other thing which I've noticed is that science always has to fulfill evidential hurdles, we always have to be transparent, we have to explain our decisions; the counter decisions don't have to go through the same level of scrutiny.

And that can result in some very strange:

this should be happening, but it's not happening, it's not happening 'cause there's an economic imperative, where's the evidence on the economics? Just trust us on that, so you get into that sort of debate. - I remember, a few years ago, a government minister discovered that half the pediatric cardiac surgeons in England were below average, and this has to stop, it just gets to the nubbins of statistical problems. Last question, I'm afraid, 'cause we're over time. - [Questioner] There are various companies that offer screening tests to test your susceptibility to disease in later life. Is it worth paying for this? - Probably not. I'll probably, now, as it's recorded online, be sued, but what they're telling you is your probability of an increased disease risk, only looking at your genetic complement, so what they're giving you is some information that you might be at greater risk. And if, for example, I had a family history of cancers of a particularly type, that would be an incredibly valuable piece of information for me to have to make sensible decisions, but if they were talking about thing such as, do you have an increased risk of chronic obstructive pulmonary disease? Well, I know the answer to your question, just don't smoke, avoid people who are smoking, if you've got a history of cardiovascular disease, there may be a genetic risk, but be lean, be fit, be healthy would be my take-home message. People want that information, they really want that information, but I think the information, sometimes, is less helpful than you think it is. - [Questioner] Thank you. - I'm very sorry to cut the questions off, we have to give the hall back, and it's very nice of you all to turn up. Thank you, Ian Mudway. (audience applauds)